Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
The terminal investment hypothesis has two predictions: in the face of an infection (i) mature males will increase investment to traits that increase mating success, while such investments will occur to a less extent in young males; and (ii) physiological costs of resource reallocation will be more severe for infected mature males than for infected young males.
Although these predictions have been tested in long-lived vertebrates, prior studies have not examined actual resource allocation conflicts. Here, we have tested the above predictions and have investigated the energetic costs of increased mating by old males, using a short-lived invertebrate, the damselfly Hetaerina americana. Males of this species defend territories as the main way to obtain access to females.
Using groups of infected vs. noninfected males of two different ages, we found that compared to young infected males, mature infected males defended territories for longer, had higher mating success and directed agonistic behaviour to conspecific males more frequently. Despite similar immune responses by mature and young males, infected mature males ended up with less fat reserves compared to infected young males. This suggests that resource allocation conflicts are more severe for mature than for young males.
In general, these results suggest that the terminal investment hypothesis applies in males of short-lived invertebrates and that a cause of increased mating success for males of advanced ages is reduced energetic stores.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
When animals reproduce several times in their lifetime, allocation of resources to current reproduction will constrain resources devoted to future reproduction (Clutton-Brock 1984; Reznick 1985; Creighton, Heflin & Belk 2009). To deal with this trade-off and thus maximize lifetime reproductive success, animals should balance resources among their expected reproductive events (Stearns 1992; Roff 2002). So, when animals perceive that they are likely to reproduce in the future or that it is not affordable to reproduce at present, they can adopt a strategy of restricting their investment to current reproduction and skip some reproductive events (e.g. Velando, Drummond & Torres 2010; Goutte et al. 2011). On the other hand, when animals perceive that the probability of mortality is high and there are low chances of reproducing in the future, they should expend higher amount of resources in current reproduction even if it ends with animal's death (Williams 1966; Javois & Tammaru 2004; Kivleniece et al. 2010). This idea is known as the ‘terminal investment hypothesis’. The predictions of the terminal investment hypothesis have been supported by studies of long-lived vertebrates (e.g. Velando, Drummond & Torres 2006; Hoffman et al. 2010) and studies of invertebrates (e.g. Adamo 1999; Javois & Tammaru 2004; Sadd et al. 2006).
The terminal investment hypothesis has two main assumptions: (i) that an investment in current reproduction is traded off against investments in future reproduction and (ii) that animals perceive their life span and thus modulate their investment in a reproductive event accordingly (Nielsen & Holman 2011). Given that prospects for future reproduction depend on individual age, the terminal investment hypothesis needs to be evaluated in individuals of different age classes, whose residual reproductive values differ (Adamo 1999; Velando, Drummond & Torres 2006). Under adverse conditions, young animals, with high residual reproductive values, should prioritize investment in self–maintenance, so that survival will not decrease drastically. On the other hand, old senescing animals, whose residual reproductive values are low, should prioritize investment in current reproduction (Forbes 1993).
A successful experimental method to probe terminal investment predictions is to reduce an animal's life expectancy via using an immune challenge in young and old animals. The immune response is a costly function (Sheldon & Verhulst 1996; Rolff & Siva-Jothy 2003) that is sensitive to both individual condition (Rantala et al. 2003; Contreras-Garduño et al. 2008) and senescence (e.g. Saino et al. 2003; Moret & Schmid-Hempel 2009). Consequently, young animals are more likely to recover from an immune insult than old animals. Terminal investment occurs when insulted animals increase their investment in reproduction when they are old but not when they are young.
Terminal investment has been mainly studied in females of both vertebrates (e.g. Hoffman et al. 2010; Weladji et al. 2010) and invertebrates (e.g. Adamo 1999; Javois & Tammaru 2004; Creighton, Heflin & Belk 2009; Cotter, Ward & Kilner 2010). In general, it has been documented that old and/or immune challenged females end up increasing investment in progeny compared to young or control females (Adamo 1999; Javois & Tammaru 2004; Creighton, Heflin & Belk 2009; Cotter, Ward & Kilner 2010; Hoffman et al. 2010 but see Weladji et al. 2010). In males that do not contribute to parental care, terminal investment can be detected as increases in the expression of sexually selected traits. Sexually selected traits are costly signals that reflect male quality and are key for male fitness (Hamilton & Zuk 1982; Anderson 1994). Several laboratory studies have investigated terminal male investment in sexually selected traits (e.g. Sadd et al. 2006; Kivleniece et al. 2010; Krams et al. 2011) and have often failed to detect shifts in male allocations (Vainikka et al. 2007). One explanation for the failure to detect this pattern is that selective pressures, such as limited food availability, that act in the wild but not under laboratory conditions are necessary to detect this pattern. A single study has been conducted in the wild, with the blue-footed booby Sula nebouxii (Velando, Drummond & Torres 2006), a long-lived bird. When old, senescing males were immune insulted, they responded by increasing their reproductive success, while younger males showed a decline in reproductive success. This study is the only experimental evidence supporting male terminal investment in the wild. Further evidence from similar wild conditions is therefore sorely needed.
An underlying assumption in studies of terminal investment is that there exists a conflict in the allocation of resources to different time-related reproductive functions (Clutton-Brock 1984; Reznick 1985; Creighton, Heflin & Belk 2009). Few studies, however, have actually assessed the energetics of resource allocation conflict. This is especially important for experimental studies as it is not at all clear, for example, whether energy devoted to immune response is subsequently not available for reproductive investment. Ignoring this energy allocation has at least two confounding effects: (i) it may be that it is not really the immune response that is costly but another noncontrolled function and (ii) the imposed challenge may not be costly at all. This may explain a few cases where the evidence of whether an energy conflict is occurring for the targeted, experimentally affected function is unclear (e.g. Langley & Clutton-Brock, 1998, Vainikka et al. 2007). An example that clearly illustrates the importance of an approach that measures energy allocation is that of the beetle Tenebrio molitor: males face an energy-based trade-off only above a threshold of immune challenge (Krams et al. 2011). Below such threshold, no energy spent in immune response was detected yet experimentally challenged males were less attractive (Kivleniece et al. 2010; Krams et al. 2011).
In the present study, we manipulated life expectancy in both young and mature males of the territorial damselfly Hetaerina americana by activating a costly immune response with a nonpathogenic elicitor (a nylon implant). We measured both aggressive territorial behaviour (used here as a sexually selected trait, cf. Contreras-Garduño et al. 2009) and mating success of implanted vs. control insects. We also measured immune response (i.e. nylon encapsulation) and energetic condition (i.e. fat reserves) after treatment. According to the terminal investment hypothesis, we predicted that mature males would show higher territorial aggression when implanted, while the opposite trend would occur in young males. We also predicted that the physiological costs of terminal investment would be higher in mature, senescent males than in young males. Specifically, we predicted that old males would lose energetic stores more drastically than young males. To our knowledge, this is the first study that experimentally evaluates the terminal investment hypothesis in males of a short-lived animal in the wild by measuring energetic costs.
Materials and methods
We used adult males of the damselfly Hetaerina americana (Odonata: Calopterygidae), which contend for riverine territories where females arrive to mate (Córdoba-Aguilar, Jiménez-Cortés & Lanz-Mendoza 2009a; Córdoba-Aguilar et al. 2009b). Males able to acquire and successfully defend territories usually have larger red wing spots and body sizes, which allow them to accrue a higher mating success compared to nonterritorial males (Grether 1996; Serrano-Meneses et al. 2007; Contreras-Garduño et al. 2008). Territorial ability, expressed in the form of aggressive flights against conspecifics, is one honest signal of male immune and energetic condition (Contreras-Garduño, Canales-Lazcano & Córdoba-Aguilar 2006; Contreras-Garduño, Lanz-Mendoza & Córdoba-Aguilar 2007; Contreras-Garduño et al. 2008). Energetic stores are present in the form of lipidic fat reserves stored in the thoracic muscle (Contreras-Garduño, Canales-Lazcano & Córdoba-Aguilar 2006). Such reserves serve as fuel during aggressive flights, and experimental work with larvae has shown that fat reserves are condition dependent (Jiménez-Cortés, Serrano-Meneses & Córdoba-Aguilar 2012). H. americana males typically have no more than three matings in their lifetime (Serrano-Meneses et al. 2007), which makes them suitable subjects for studying senescence in reproductive activity and terminal investment.
Field work was carried out in the Apatlaco river, Morelos, Mexico (18°45′55″N, 99°14′45″W), from November to December 2011, and in Metztitlán river, Hidalgo, Mexico (20°32′30″N, 98°43′45″W), from December 2003 to February 2004. Animals were captured with a butterfly net between 1100 and 1400 hours, the time at which territorial activity is highest at our study site (all authors' personal observations). Behavioural observations were also carried out in this time interval. Given that our main aim is to study senescence, we took special care to assess male age. We classified ages in categories according to visible features of the wings, thorax and abdomen (Plaistow & Siva-Jothy 1996): age 1 males have soft, undamaged and dorsoventrally flexible wings; age 2 males are young and sexually mature individuals that fight for territories, have harder wings that are flexible from the nodus to the tip and are already pigmented; age 3 males are fully sexually mature but have less flexible wings and show some signs of pruinescence in the thorax; age 4 males, the oldest, show abundant thoracic pruinescence and inflexible wings that are occasionally damaged (broken at their tips). We were also careful about territorial status and considered that a male was territorial when it was observed involved in contests with conspecific males and actively defending a site. Territorial males also reacted aggressively to the presentation of an experimental conspecific intruder. Otherwise, males were considered nonterritorial. We only collected territorial, sexually mature males of ages 2 and 3. Based on the age categories mentioned above, we created two experimental categories: ‘young’ and ‘mature’. Males were considered ‘young’ when their wings, thorax and abdomen were still brilliant and the pterostigmata were still clear. All young animals were of age class 2. Some males of age 3 were marked with a small dot in the right anterior wing, released and recaptured 8–10 days later only if they were territorial at that time. These are ‘mature’ males that we estimate to be at least 10–15 days older than young males. Marked males that were already from age class 4 were excluded from the study. We considered that an age difference of 10–15 days was appropriate for our experiment given that the longevity of these animals is around 30 days (all authors' unpublished data) and at the age of 15 days since emergence, damselflies of similar life span can start showing senescence (Sherratt et al. 2010). While our ageing method is based on subjective traits (appearance of the animals), the fact that young and mature males differed in body size in our experiments suggests that they emerged at different times and thus come from different but overlapping cohorts (see also Córdoba-Aguilar 2009a).
Both young and mature males were manipulated with either an experimental or a sham treatment. In the experimental group, an immune response was triggered by inserting a nylon monofilament (previously rubbed with fine sandpaper, 2 mm long, 0·18 mm width) in the ventral part of the fourth abdominal segment (Rantala et al. 2000). When implanted, insects respond by melanizing the implant (for similar methods, see Rantala & Roff 2007; González-Tokman, Córdoba-Aguilar & Forbes 2012). Males in the sham group received the same manipulation, but the implant was immediately removed. Males were anaesthetized in ice for 10 min, marked with a unique number made with permanent ink on the right anterior wing (numbers can be seen from 3 to 5 m away) and released to the same place where they were collected.
Experiment 1: Long-term territorial defence and mating success
In the Metztitlán river, we collected and manipulated 51 experimental young males, 59 sham young males, 56 experimental mature males and 73 sham mature males. Wing length, measured from the site of wing insertion in the thorax to the distal end of the wing, was considered as the measure of body size (Serrano-Meneses et al. 2007). Starting 24 h after manipulation (experimental and sham), animals were tracked daily to record territory tenancy and mating success for 19–20 days. For territory tenancy, we walked three times daily along the river stretch (c. 200 m length) where marked males were released and recorded which males were present on each site. Sites were depicted on a map of the river stretch and were updated daily (as defended areas may move depending on sun conditions; Córdoba-Aguilar et al. 2009b; Córdoba-Aguilar, Jiménez-Cortés & Lanz-Mendoza 2009a; see also Córdoba-Aguilar 1995). Unlike males of the experiment 2 described below, territory tenancy was estimated by seeing whether males remained at the same site in at least on two of the three surveys. We did not directly experimentally assay territoriality as there was too much vegetation in the Metztitlán river to simulate intrusions as in experiment 2. Despite this, we are confident that this approach was useful to record territory tenancy as males that failed to appear on a particular day at their previously recorded site did not subsequently reappear on the same spot and actually wandered to different sites. This indicates that males became nonterritorial after losing a territory (see also Raihani, Serrano-Meneses & Córdoba-Aguilar 2008). Mating success was also recorded via these daily censuses. Although matings are relatively short (2–3 min), the fact that a copulating pair usually flies continuously for up to 40 min to reduce male harassment makes visually recording each mating relatively tractable (Córdoba-Aguilar 2009b).
Experiment 2: Short-term male aggressiveness and physiological condition
In the Tetlama river, we collected and manipulated 52 experimental young males, 48 sham young males, 52 experimental mature males and 53 sham mature males. Twenty-four hours after manipulation (nylon insertion or sham treatment), we made behavioural observations and recaptured the animals to recover the implants and measure the extent of melanization. Animals were then sacrificed and stored in 70% ethanol for subsequent measurement of fat content. We took care to observe young and mature animals simultaneously, so differences in behaviour or physiology between age classes were not due to climatic conditions. Given that within odonates it is common that time of emergence affects morphological features (Stoks & Córdoba-Aguilar 2012), there were unavoidable differences in the body sizes of young and mature males in our study (see 'Results'). We addressed this issue by including body size as covariate in statistical analyses. Some animals could not be recaptured after observation and for some of them implants could not be recovered, so sample sizes may differ among different analyses.
Experiment 2: Behaviour
Each focal male's territorial behaviour was recorded in response to simulated intrusions of conspecific males. Intruding conspecifics were fully mature males of age class 3 that were captured at the moment of observation and tethered to a 50-cm nylon thread attached to a stick (Anderson & Grether 2010). Tethered intruders were presented ten times to each focal male, and we waited 2 min between intrusions. A presentation was considered complete when the tethered intruder flew for at least 5 s not further than 20 cm from the front or side of the focal male. Tethered intruders were replaced when they stopped flying, so that in some cases, more than one was used per focal male or sometimes one intruder was used for more than one focal male. For each presentation, we classified focal behaviour as one of three possible behavioural responses: (i) a focal male was considered to ‘attack’ the model when he responded to the presentation by flying towards and chasing the intruder. This is the clearest territorial response (Anderson & Grether 2010); (ii) a male was considered to do a ‘wing display’ when it opened its four wings without moving from its site. While this behaviour has not been considered as a territorial display in Hetaerina damselflies (Anderson & Grether 2010), it is used to intimidate males in odonates (Utzeri 1988). Finally, (iii) a male was considered to ‘getaway’ if it flew at least 2 m apart of its site when the intruder was presented (see also Córdoba-Aguilar 1995). We only considered individuals with at least eight recordings. Sample sizes were as follows: young implanted N = 18, young sham N = 19, mature implanted N = 22 and mature sham N = 26.
Experiment 2: Immune response, fat content and body size
Melanization of nylon implants was measured as the darkness of an implant relative to a control piece of nylon that was not implanted. Each implant was photographed from three different angles at one side of the control, and darkness was obtained with adobe photoshop 7.0 (Adobe Systems Inc., San Francisco, CA, USA), using a scale where 0 is the minimum possible value and means black, and 255 is the maximum and means white. The average from the three pictures of the experimental nylon divided by the average of three pictures of an uninserted nylon was considered implant darkness. Sample sizes for melanization of nylon implants remained as follows: young N = 14 and mature N = 21.
For measurements of fat content, animals were dried in a desiccator, weighed (±0·1 mg), submerged in chloroform for 24 h for fat extraction, redesiccated and reweighed (for similar procedures, see Marden 1989; Pekár et al. 2010; González-Tokman, Córdoba-Aguilar & Forbes 2012). The difference between the initial and the final weights was considered fat content. Sample sizes for fat content were as follows: young implanted N = 19, young sham N = 19, mature implanted N = 21 and mature sham N = 22.
As a measure of body size, we used the area of the right anterior wing (in mm2). We used wing area instead of wing length because in experiment 2, we photographed males with a scale of known area and did not measure wing length, as it has been done in experiment 1. To measure wing area, we analysed pictures of the animals with adobe photoshop 7.0.
Initial differences in body size between males of different age classes and treatments were tested with two-way analysis of variance (anova), with age and treatment as factors. Given that body size was different between young and mature males in both the long- and the short-term experiments (see 'Results' section), body size was always included as a covariate in subsequent analyses.
In experiment 1 (long term), the number of days that males defended a territory and the probability of copulating were analysed with Generalized Linear Models (GLM). The initial models tested included age, treatment, body size and all interactions (age × treatment × body size). Model selection was carried out both backwards and forward based on AIC values of all competing models (Johnson & Omland 2004). Overdispersion (residual deviance/residual d.f.) was low (< 2) in all global models. When the response variable was the proportion of days where focal males defended their territory, we used a GLM with binomial distribution of errors and logit link function. When the response variable was the number of copulations, we used a GLM with Poisson distribution of errors and log link function.
In experiment 2 (short term), male behaviour in response to simulated intrusions of conspecific males was also analysed with GLM. The initial models tested included age, treatment, body size and all interactions (age * treatment * body size). As above, model selection was based on AICs. Overdispersion was tested in the global models and was low (< 2) unless otherwise specified. When the response variable was the proportion of events where the focal male attacked the intruder, overdispersion of the global model was high (residual deviance/residual d.f. = 6·55), and then, we used quasi-binomial errors and logit link function. Starting with nonsignificant interactions, less significant covariates were eliminated one by one from the initial model based on the significance of a likelihood ratio tests (LRT) between the initial model and the model without the covariate. A significant LRT (P < 0·05) means that the exclusion of the covariate does not affect model fit, and then, the covariate can be removed from the analysis (Johnson & Omland 2004). When the response variable was the proportion of intrusions where the focal male showed his wings without flying, we used a GLM with binomial errors and logit link function. Including the effect of heterogeneous variances in males of different age and treatment was not significant (beta-binomial model LRT age: P =0·48, treatment: P =0·37). When the response variable was the proportion of intrusions where the focal male left his perch, we used a GLM with binomial errors and logit link function. Including the effect of heterogeneous variances in males of different age was not significant (LRT between beta-binomial models P =0·66).
Melanization of nylon implants and fat content were analysed with general linear models (LM). Model selection was performed backwards and forward based on AIC values and LRT. When implant darkness was the response variable, the initial LM included age, body size and the interaction (age * body size). When fat content was the response variable, the initial LM included age, treatment, body size and all interactions (age × treatment × body size). Homogeneity of variances was tested with Fligner–Killeen tests.
In the 'Results' section, we only show the best supported models. The presence of outliers was tested with Cook's distance, but no outliers or influential points were detected in any analysis (all Cook's distances < 0·5; Crawley 2007). Analyses were carried out in R version 2.10.0 (R Development Core Team 2009).
Experiment 1: Long-term territorial defence and mating success
Young males were significantly larger than mature males (F1,236 = 12·42, P <0·01, N = 239; mature: 25·40 ±0·05 mm, N = 129; young: 25·70 ± 0·07 mm, N = 110). There were no differences in male body size between treatments (F1,236 = 0·32, P =0·57).
The probability of holding a territory depended on the interaction of age and treatment and on the interaction of body size and treatment (Figs 1 and 2; binomial GLM d.f. = 233, age z =1·17, P =0·24; treatment z =1·55, P =0·12; body size z =3·37, P < 0·01; age: treatment z =7·32, P ˂ 0·01; treatment: body size z =−1·95, P =0·05). Young males lost their ability to defend a territory when implanted, while no effect of implant was found in mature males (Fig. 1). Also, larger males remained territorial for more days in the sham treatment but not in the implanted treatment (Fig. 2). The number of copulations was higher in mature and large males (Poisson GLM d.f. = 236, age z =3·64, P <0·01; body size z =3·88, P <0·01), independently of treatment (Fig. 3).
Experiment 2: Short-term male aggressiveness and physiological condition
Contrary to the long-term experiment, mature males were significantly larger than young males (ancova F1,199 = 12·33, P < 0·001; mature: 115·75 ± 8·97 mm2, N = 104; young: 110·95 ± 10·41 mm2, N = 98). There were no differences in male body size between treatments (F1,199 = 0·08, P =0·78).
Male aggressiveness, measured as the proportion of encounters where a focal male attacked the conspecific intruder, was not dependent on any of the tested covariates (and interactions): age (young/old), treatment (implanted/control) and body size (quasi-binomial GLM, Table S1).
The proportion of encounters where the focal male opened his wings (wing display) in response to the intruder was marginally dependent on the interaction of age and treatment (binomial GLM d.f. = 80, age: z =2·89, P <0·01; treatment: z =1·08, P =0·28; age: treatment: z =−1·92, P =0·06), with mature males displaying more intensively when infected and young males showing the opposite trend (Fig. 4).
The proportion of encounters in which the focal male left his place in the presence of an intruder was marginally dependent on the interaction of age and body size (binomial GLM d.f. = 80, age z =−1·85, P =0·07; body size z =−0·58, P =0·56; age: body size z =1·71, P =0·09). The trend was that young males left the sites more often when they were smaller, while mature males left their sites more often when they were larger (Fig. 5). In general, young males left their site more often than mature males, and this effect was independent of treatment.
Melanization of nylon implants was not dependent on age or body size (LM; Table S2). The amount of fat reserves in male's thoraces was dependent on body size and on the interaction age: treatment (LM d.f. = 76; age t =1·87, P =0·07; treatment t =1·80, P =0·08, body size t =2·24, P =0·03; age: treatment t =2·22, P =0·03). Larger males had more fat reserves. Young animals had more fat reserves when implanted, and old animals had less fat reserves when implanted (Fig. 6).
When animals get old, they adjust their investment in reproduction in two possible ways: first, old animals can increase their investment in current reproduction as a strategy of terminal investment if they perceive that their chances of future reproduction (residual reproductive value) are reduced (Clutton-Brock 1984; Velando, Drummond & Torres 2006). Alternatively, old animals can reduce their investment in current reproduction either because they end up deteriorated as a result of senescence or because of an adaptive reproductive restraint (McNamara et al. 2009). Here, we have used male H. americana damselflies to demonstrate that when chances for future reproduction are experimentally reduced by an immune challenge, old animals show terminal investment because they maintain elevated territorial activity despite a deteriorated physical condition. Our results imply that damselflies somehow perceive their future chances for reproduction and allocate their resources accordingly. Unlike other systems (Velando, Drummond & Torres 2006; Cotter, Ward & Kilner 2010), old H. americana males do not seem to be cautious about their investment in reproduction, and as a consequence, their energetic reserves become depleted.
To evaluate our first prediction – that mature implanted males would have higher mating success than young implanted males – we found mixed support. Mature males indeed did accrue higher mating success than young males; however, this effect was independent of treatment. In our study species, territorial males rarely get more than three matings in their life, while most nonterritorial males obtain no matings at all (Serrano-Meneses et al. 2007). Furthermore, only 10–15% territorial males obtain some matings (Serrano-Meneses et al. 2007). Such strict characteristics make the Hetaerina mating system an extremely competitive biological system. Thus, that implanted mature males invest their energy and time to still hold a territory (and so accrue some matings) can be interpreted as a strategy of terminal investment.
We predicted that higher mating success in mature implanted males would be linked to enhanced territorial aggression. However, we did not find an effect of age or treatment in aggressiveness towards intruder males. Instead, our evidence, although marginally significant (P <0·10), suggests that there is an increased frequency of wing displays in mature implanted males and that mature males are less likely to abandon their territories. Why would wing displays be functional in the context of territorial defence? In odonates, these displays are considered as threat signals that males and females use against any sex of the same or different species (Corbet 1999). These type of displays in territorial animals can be used as alerting signals that prevent direct confrontation between males (Maynard-Smith & Harper 2003; Searcy & Nowicki 2005; Grether 2011). We suggest that H. americana wing display by males has the same function and that, when used against conspecific males, is to show the red spot males have on at the base of their wings. Previous studies have indicated that the size of such spot is a condition-dependent trait (Contreras-Garduño et al. 2008; Jiménez-Cortés, Serrano-Meneses & Córdoba-Aguilar 2012). Displaying this trait in the presence of intruder males could be a way to avoid confrontations especially when the territory holder is sick, as it was the case with implanted animals. Interestingly, even when mature implanted males had less fat reserves (the prime fuel used during odonate flying contests; Marden 2008) than young implanted males, mature males still pursued a more aggressive strategy to hold their territories. This can be explained in terms of the fitness pay-offs for mature males as these had more to lose than young males.
Our study is not the first to examine the effect of activation of the immune response on territoriality in the same and closely related species. In H. americana, prior studies have found that mature males infected with bacteria defended their territories with the same intensity as healthy mature males (González-Tokman et al. 2011). In young males, on the other hand, the probability of becoming territorial is lower for animals that were immune challenged in closely related species of the genus Calopteryx (Rantala, Honkavaara & Suhonen 2010). This probably occurs because young infected individuals disperse further to get new territories, to feed better or to avoid reinfection (Suhonen, Honkavaara & Rantala 2010).
We believe that losing a territory rather than defending it can be an adaptive strategy used by young males to increase their fitness, especially when infected. Previous research in H. americana has found that despite the fact that being nonterritorial means nearly zero matings, there are two options to gain at least one mating. First, if a nonterritorial male is large enough, he can regain a territory (Raihani, Serrano-Meneses & Córdoba-Aguilar 2008). Of course, this option depends on whether the male has enough energetic resources in the form of fat reserves (Raihani, Serrano-Meneses & Córdoba-Aguilar 2008), which seems the case of our infected young males. A second option, which produces a reduced fitness outcome, is to wander over several territories as a nonterritorial male and opportunistically take over flying mating couples to displace the mating male (Córdoba-Aguilar et al. 2009b). This strategy can be effective only if the nonterritorial male is larger than territorial males, which is not that common in odonates (Suhonen, Rantala & Honkavaara 2008).
Prior studies using immune challenges to modify animal condition and thus test predictions from the terminal investment hypothesis have not looked at potential resource allocation costs in energetic terms. In our work, we examined allocation costs in energetic terms by measuring thoracic fat reserves. We found that melanization of the nylon implants did not differ according to age. This means that males of different ages invest the same in immune response, but despite this, infected mature males suffered increased losses to fat reserves. A reduction in fat reserves at any age can have varied negative effects. One negative effect is reduced survival (Contreras-Garduño, Lanz-Mendoza & Córdoba-Aguilar 2007). In functional terms, the reduction in fat stores in mature males may be partially explained by a resource reallocation from immunity to fat reserves, which affected mature males more than young males. Such resource reallocation cannot be confused with the cost of territorial defence as the rate of territorial attacks to intruders was the same for all experimental groups.
Young infected males may have engaged in compensatory resource intake. This would account for their ability to maintain energetic reserves following the immune challenge. Compensatory resource intake has been observed in the same species (González-Tokman et al. 2011) and other insects (Lee et al. 2006; Povey et al. 2009) and is used to compensate an energetic imbalance when faced with energetic problems such as an infection. Old infected males did not compensate probably because either they invested their remaining time to get more matings or they were physiologically incapable of transforming more food into energetic reserves (e.g. Siva-Jothy & Plaistow 1999). However, previous results in this species indicate that even relatively old animals are able to restore fat reserves, which suggest that feeding can occur at different ages and not only young ones (Raihani, Serrano-Meneses & Córdoba-Aguilar 2008).
Body size showed different trends in both experiments, which can be partially understood on the basis of effects of seasonality that have been detected in this (Córdoba-Aguilar 2009a) and other insect and odonate species (Forrest 1987; Kause et al. 2001; Stoks & Córdoba-Aguilar 2012). According to such effects, body size tends to change along the season in H. americana. The fact that both experiments were carried out in relatively different times and sites may explain such differences in body size.
Given that our study was carried out in wild animals under natural conditions, there are some factors that escaped our control. Unlike studies in laboratory reared animals, we could not have absolute control of initial male age, feeding intensity or mating experience. Our ageing method guaranteed that mature males were at least 15 days older than young males. However, young and mature males emerged at different times (and therefore they differed in body size), so knowing the precise age would have been desirable. Moreover, we have argued that compensatory feeding could have occurred in our wild population, but we did not observe nor have control of foraging behaviour. Also, mating experience could have caused differences in behaviour and condition that we could not control. Although studies in captivity can control all these factors, they not always represent real situations faced by wild animals and that is why studies in wild animals need to be complemented with studies in captivity. Despite the heterogeneity in the initial conditions of the animals we tested, we still found clear trends supporting terminal investment hypothesis in males of a territorial insect.
The terminal investment hypothesis in short-lived animals has been poorly studied, especially in males. Our results show that this hypothesis applies to males of a short-lived invertebrate. In H. americana, young infected males seem to take the option of leaving a territory and possibly wait for more favourable conditions, while old infected males prefer to stay in a territory and consume their reduced energetic stores to get as many matings as noninfected males.
The authors would like to thank PAPIIT project Nos. IN 204610 and IN 222312-3 for financial support, H. Hernández-Córdoba and F. Baena-Díaz for help during fieldwork and R. Munguía-Steyer for help on statistical analyses. A. Velando and C. N. Anderson provided valuable comments on the manuscript. This paper constitutes a partial fulfilment of the Graduate Program in Biological Sciences of the Universidad Nacional Autónoma de México. DG-T and IG-S acknowledge the scholarship provided by the Consejo Nacional de Ciencia y Tecnología (CONACyT, México). All authors confirm not to have any conflict of interest. Damselfly illustrations are by Barrett Klein, originally published in Abbott (2011).